Upgrade from 8.8.12 to 8.8.15 has gone south. Please assist.

[tgx]

Upgrade went fine without any errors but service is offline due to zimbra proxy which is in status stopped.
After shutting down all services and trying to restart I get the message popped:

Starting proxy...nginx: [emerg] host not found /opt/zimbra/conf/nginx/includes/nginx.conf.web:27 failed

Looking in that file I see a lot of references to a servername that is no longer in use. The server was renamed
quite some time ago. Also a lot of references to the previous version of 8.8.12..
eg:

Code: Select all

upstream zimbra_webclient
    {
        server    oldservername.domain.com:8080 fail_timeout=10s version=8.8.12_GA_3794;

        zmauth;
    }

Any help appreciated.

[tgx]

Update*** I have been able to get the service to come up by adding in the older servername into /etc/hosts.
How should I proceed? Should I run through the steps listed here:

http://www.aionsolution.com/it-support/ ... ervername/

to set the servername again? Apparently it only partly took the last time.

[tgx]

You can see the places where oldservername is still in effect.

[zimbra@servername libexec]$ zmprov gacf | grep oldservername.domain.com
zimbraReverseProxyAvailableLookupTargets: oldservername.domain.com
zimbraReverseProxyUpstreamLoginServers: oldservername.domain.com
zimbraVersionCheckNotificationEmail: user@oldservername.domain.com
zimbraVersionCheckNotificationEmailFrom: user@oldservername.domain.com

[zimbra@servername libexec]$ zmprov gacf | grep servername.domain.com
zimbraAmavisQuarantineAccount: virus-quarantine@servername.domain.com
zimbraDefaultDomainName: servername.domain.com
zimbraLogHostname: servername.domain.com
zimbraSpamIsNotSpamAccount: ham@servername.domain.com
zimbraSpamIsSpamAccount: spam@servername.domain.com

[zimbra@servername libexec]$ zmprov gs 'servername.domain.com' | grep oldservername.domain.com
zimbraReverseProxyAvailableLookupTargets: oldservername.domain.com
zimbraReverseProxyUpstreamLoginServers: oldservername.domain.com

[tgx]

This issue is still open if anyone would like to assist.
Thanks.

[phoenix]

What about the answer I gave you on the Zextras forums?

[tgx]

What about the answer I gave you on the Zextras forums?

You mean this one?

Sometime you don't get an answer because people might assume you'd try changing those attributes individually or even try changing the server name again to the 'new' name and see if that fixes it. In any case, your question is about ZCS and I'd guess it's not suitable for these forums as it's not a Zextras problem/question. "

I do not have knowledge of how best to change the attributes, if I did the question would not have been posted in a forum.
Nor do I even know if simply changing the attributes is the correct approach to take. Also, I do not know if my original process for setting the servername was correct
as there is no consistent online source of information. I felt asking in a forum may provide the most up to date approach to solving this issue.

[tgx]

The original setservername process I followed was here, dated 2017:

https://wiki.zimbra.com/wiki/ZmSetServerName

I just reran this portion because there were oldservername entries in there.
Also I now see one for 'localhost' which was not previously in there prior to the upgrade.
Also inside Zimbra adminconsole I have monitor entries for my servername AND localhost with
exception that all localhost entries are showing red X's.

**Looks like you can get rid of localhost.
https://wiki.zimbra.com/wiki/Server_sta ... rossed_out

These are the commands I have run to clear that up. I cannot bounce my mailboxd at this time
to test these changes as I have users working.

Code: Select all

zmloggerhostmap -d oldservename oldservername.domain.com
zmloggerhostmap -d oldservename.domain.com oldservername.domain.com

zmloggerhostmap

Zimbra Logger Service Hostname Map

---------------------    ---------------------
DNS Hostname             Zimbra Hostname
---------------------    ---------------------
servername servername.domain.com
servername.domain.com servername.domain.com
localhost localhost

I don't like messing with LDAP as it's a quick road to a dead server.

I am thinking my next steps may be here, but I am uncertain.

Code: Select all

Only place valid mailbox server names in these attributes. If there are non-existing mailbox servers (like an MTA, or a decommissioned server) change the value:

zmprov ms server_name zimbraReverseProxyUpstreamEwsServers <mailbox_server_name>
zmprov mcf zimbraReverseProxyUpstreamEwsServers <mailbox_server_name>

Then run zmproxyconfgen:

/opt/zimbra/libexec/zmproxyconfgen -s `zmhostname`

Those suggestions came from here after googling 'zimbra zimbraReverseProxyAvailableLookupTarget'

https://wiki.zimbra.com/wiki/Nginx_fails_to_start

[tgx]

Okay, so in summary I have issued:

Code: Select all

zmprov mcf zimbraReverseProxyAvailableLookupTargets servername.domain.com
zmprov mcf zimbraReverseProxyUpstreamLoginServers servername.domain.com
zmprov mcf zimbraVersionCheckNotificationEmail user@servername.domian.com
zmprov mcf zimbraVersionCheckNotificationEmailFrom user@servername.domain.com
zmloggerhostmap -d localhost localhost
zmloggerhostmap -d oldservername oldservername.domain.com
zmloggerhostmap -d oldservername.domain.com oldservername.domain.com

I also ran the equivalent zmprov gs and zmprov gacf commands to check that the changes have been applied.
All that is left is to bounce services and learn what I missed.

I guess one remaining question is if this needs to be run or will it execute it automagically at service run time:

Code: Select all

/opt/zimbra/libexec/zmproxyconfgen -s `zmhostname`

[tgx]

All is well after the service bounce.